The console saga continues

Martin Mares (
Sun, 26 Jul 1998 16:16:09 +0200

Hello, world!

After another week of hacking, the console seems to be working now and my bug
list is empty (famous last words ;-]). The only things I'd want to do are:

- MDA console for dual-head machines
- Stripped down `bootup VGA console' for people using X and frame buffers.
- Move include files from drivers/video to include/video.

You can find the patch against 2.1.111 at

Summary of changes:

Martin Mares [mm]
Jakub Jelinke [jj]
Geert Uytterhoeven [ge]
Eddie C. Dost [ed]

console and vt:
o Checks for existence of console we're switching to moved from change_console()
to console_bh(), so that the cursor doesn't disappear when trying to switch
to non-existent console. [mm]
o do_con_write(): Clear cursor _after_ clearing selection, so that the scrollback
doesn't get reset too soon on vgacon. [mm]
o Make unimaps a per-console thing, but share it as much as possible.
For vgacon & promcon, unimap changes change all consoles (as they
share fonts), otherwise just the current one. [jj]
o Fix GIO_FONT/GIO_FONTX/KDFONTOP, finally I can get fonts from kernel. [jj]
o Allow chardata/data to be NULL in GIO_FONT* and KDFONTOP - some programs
rely on it to get just attributes of the font. [jj]
o Move the hi_font_mask/build_attr stuff from fbcon into generic code, plus fix it. [jj]
o Allow font loading/retrieval for wide fonts as well (you need the patched
consoletools; I have just posted the patch to the list). [jj]
o Compile kd_mksound only on architectures which need it. Also updated
few obsolete comments around. [mm]
o Removed few inlines [mm]
o Screen alignment test works again. [mm]
o clear_selection() called from set/hide_cursor instead of random spots
in the code. Selection now gets always cleared _before_ resetting
scrollback. [mm]
o Don't attempt to back up attributes under the cursor if softcursor
is disabled. [mm]
o Fixed APM unblanking. [mm]

o vcs scrbuf access functions moved to console.c and they are
softcursor-corrected now. [mm]

Sparc framebuffers:
TCX, BW2, CG3 support. Please test if you can and report any problems to me. [jj]

Extensive changes [ge] [ed]

Mac framebuffers:
o New frame buffer device for PowerMac `Control' display adaptor.
(by Daniel Jacobowitz <>)
o Frame buffer device for the PowerMac 'platinum' display by Dan Jacobowitz

o Add support for fontwidths {4,12,16} to fbcon-cfb{16,24,32}. [ge]
o Implement clear_margins() for fbcon-cfb{16,24,32}. [ge]
o Add clear_margins() calls to fbcon_aty{16,32}. [ge]
o Support 512ch fonts [jj]
o A config option to turn off support for widths != 8. [jj]
o If you load the same font on a couple of consoles, share the fontdata, don't
waste memory. [jj]
o If logo is shown, set scroll region top below it, so it does not scroll away. [jj]

Disable caching correctly on i386. [mm]

o promcon: by default it won't come up (use console=prom), but
if no fbs are probed, it will take_over [jj]
o promcon: use a modified latin1 unimap with emulated linedrawing chars [jj]

Experimental frame buffer device for Amiga boards based on the CL-GD542x/543x
by Frank Neumann <> and Peter Remmers
<>. This driver uses fbgen.
o Updated for the latest kernel version, some clean ups. [ge]
o Moved fbgen prototypes to <linux/fb.h>. [ge]
o Fixed colormap bugs for `high' color depths. [ge]

o Don't make dummycon too big on sparc, the logo insertion logic
has problems with going from large to small console... [ed]

Have a nice fortnight

Martin `MJ' Mares   <>
Faculty of Math and Physics, Charles University, Prague, Czech Rep., Earth
"Purchasing Windows is an Unrecoverable Application Error."

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to Please read the FAQ at